Ralf Stuckert created SUREFIRE-897:
--------------------------------------
Summary: System.exit() in ForkedBooter might hang due to
swing/windows bug
Key: SUREFIRE-897
URL: https://jira.codehaus.org/browse/SUREFIRE-897
Project: Maven Surefire
Issue Type: Bug
Components: process forking
Affects Versions: 2.12
Environment: Windows, Java 6,7
Reporter: Ralf Stuckert
Attachments: ForkedBooter.patch
Due to a bug in Swing on Windows (see
http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=7071160) the forked process
may not respond to System.exit() in class ForkedBooter after excuting swing
tests on windows. This leads to a hanging build, and the only way to resolve
it, is to kill the hanging process.
After all, sending Runtime.halt() will always stop the process. But since this
is not a clean shutdown (e.g. shutdown-hooks not running), this should not be
the default behaviour.
As a workaround, you could start a daemon thread as a watchdog before calling
System.exit().
The deamon sleeps for a certain time, let's say a minute. If it ever resumes
from sleeping, it means that the JVM is still running, so the System.exit() was
not completed in that time. Now it's time to call Runtime.halt() as a last exit.
We used this strategy to overcome the Swing bug (see attached patch).
I guess it would be nice, if the timeout would be configurable.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://jira.codehaus.org/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
